Today, Tailscale has announced new capabilities for Aperture, its AI access and control platform, designed to provide IT teams with a common and stable layer for managing AI across evolving models, tools, and data sources.

The shadow AI problem

AI tools are already being used inside most companies, whether the IT department officially supports them or not. The problem is that this usage is often invisible, siloed, and extremely difficult to secure. Employees often use free personal accounts, different teams adopt different tools with credit card payments, and agents are beginning to act within systems originally built and designed for people.

According to recent research mentioned by Tailscale, over 64% of activity on personal and free AI accounts is for work. This creates a blind spot for IT teams who cannot see, govern, or recover that data. Other research found that companies typically have nearly 70 generative AI tools running across their systems, with 90% lacking proper licensing and/or approval.

AI providers are heavily incentivized to bundle models, chat interfaces, and execution environments into closed stacks. While those bundles can make the initial rollout for organizations easier, they introduce vendor lock-in, tying you to a single vendor.

How Aperture aims to fix this

Aperture aims to give organizations an easier way to manage AI without locking them into a single vendor. It makes approved AI tools easier to use and provides agents with controlled environments in which to work. Just as importantly, it keeps the AI stack modular.

• Chat interface: This gives employees a browser-based way to use approved AI models. The interface supports switching between configured LLM providers as well
• Universal data connectors: These help AI tools reach internal tools, documents, and operational data without forcing every single team to build its own integration.
• Identity preserved across systems: Because Aperture integrates cleanly with Tailscale, user identity is preserved, and permissions are carried through the entire agent lifecycle.
• Sandbox support: Available in private alpha, this provides AI agents with controlled environments in which they can operate without directly acting on a user’s local laptop or an unmanaged system.

Aperture is designed to work with API keys from major LLM providers, including OpenAI, Anthropic, Google Gemini, and Amazon Bedrock. The interface and universal data connectors are available today in public alpha for organizations currently using Aperture.

9to5Mac’s take

The AI stack is not going to settle down any time soon. News by the week changes your approach. The best model, interface, and data connection will keep changing rapidly. Companies should not have to rebuild their AI setup every time one of those pieces changes, or they want to mix in a new vendor.

Aperture aims to provide IT departments with a stable layer for identity, access, and control, so they can keep changing tools without losing track of who is doing what in the corporate environment. Moving AI agents into a secure sandbox instead of letting them run wild on a corporate Mac is exactly the kind of control IT needs right now.